Tuesday, August 30, 2005

 

GP and Intelligent Design

Genetic programmer, Lee Spector of Hampshire College, has an op-ed piece in the Boston Globe in which he talks about evolving quantum algorithms and intelligent design.

Monday, August 29, 2005

 

Autonomous flying robots

Earlier this year, Dave Goldberg blogged about autonomous choppers. I came across a recent post at ai+alife+aculture describing bio-inspired vision-based flying robots (some stages of the work are GA-based). The work is a Ph.D project of Jean-Christophe Zufferey headed by Prof. Dario Floreano in Switzerland.

The researchers say
In this project, we explore an approach whereby robust vision-based behaviors emerge out of the coordination of several visuo-motor components that can directly link simple visual features to motor commands. Biological inspiration is taken from insect vision and evolutionary algorithms are used to evolve efficient neural networks. The resulting controllers select, develop, and exploit visuo-motor components that are tailored to the information relevant for the particular environment, robot morphology, and behavior.


A cool QuickTime video of the flight can be seen here. Close-ups of the plane are shown a little more than halfway through the clip.

Sunday, August 28, 2005

 

Business vs. engineering ethics?

MarkN has an interesting comment on business ethics versus engineering ethics, suggesting from personal experience that the former is more problematic than the latter. He also suggests possible causes for the observed difference:
I often wonder what it is about business education versus engineering education that turns out such diametrically opposed (on average) world views. Or are there a set of personality characteristics that defines both education preferences and ethical behavior preferences?

I wonder whether there is data to support MarkN's basic assertion. If his assertion is true, the notions that education and personality differences might help explain the difference sound plausible. Also, could it be that the role of businesspeople places them in environments richer in potential conflicts of interest and ethical dilemmas than engineers. Moreover, could it be that the potential rewards from ethical misconduct are greater for the businessperson. Either way, some straightforward reasoning shows how these factors might explain MarkN's observation.

Assume a probability p of successful ethical response to an ethical dilemma. Assuming a constant p across the population. If businesspeople face more dilemmas, then they will have a greater expected number of violations.

Assume that the probability p is not constant. Instead assume that success in facing an ethical dilemma is a decreasing function of the magnitude of reward available for an ethical violation. In an environment with a constant number of opportunities for ethical misdeeds, but where the engineer faces less reward than the businessperson for his or her misconduct, the business community will be observed to have a higher per capita incidence of misconduct.

Perhaps, in this light, MarkN's comment is not so surprising. In the future, as engineers become more entrepreneurial, as I believe they are becoming, we can probably expect greater opportunities and rewards for ethical misconduct to make it harder for engineers to choose to do the right thing.

 

Virtual Journal tags GP paper

The Virtual Journal of Nanoscale Science & Technology has chosen the article "Genetic programming for multitimescale modeling" (Sastry et al., 2005, Physical Review B 72, 085438) for the 8/29/05 edition under the subheading of surface and interface properties. The article is the product of NSF and AFOSR research in using genetic programming and genetic algorithms to bridge calculations at different timescales. The work is part of IlliGAL blogger Kumara Sastry's PhD dissertation research. Kumara is jointly advised by Duane Johnson (MatSE) and Dave Goldberg (General Engineering).

 

Pi-rats is blogging GAs

here. The first post has a short layperson's description of genetic algorithms.

 

Here be dragons

A white paper out of USC outlines the creation of an adaptive gaming environment using L-systems and genetic algorithms. Its not clear how much of this working yet, and its not clear how much of this is new. Although commercial coders don't usually reveal their underlying technique, certainly projects like Spore use techniques at least as effective as those outlined here. Nonetheless, looking under the hood is valuable, and if nothing else, the author has chosen a name that captures bloggers' attention. (via occult design)

Thursday, August 25, 2005

 

The Entrepreneurial Engineer is coming

The long-awaited [:-)] rewrite of my 1995 book, Life Skills and Leadership for Engineers, is nearing completion and a short course covering the same material will soon be available (see here). The book, now entitled The Entrepreneurial Engineer, has undergone major revision, including major additions to the chapters on writing, engagement, teamwork, and leadership. New chapters have been added on ethics and technology opportunity assessment. Stay tuned for further announcements.

 

Still time to sign up for GE 531

Classes began yesterday at the University of Illinois, and I'm teaching my genetic algorithms course again after a one year layoff due to my sabbatical. The course is now numbered GE 531 and the web site is here. The course may be taken online for UIUC credit and Laura Miller (arriola@uiuc.edu or phone: 1-800-252-1360 x36634) can provide registration information. As of this morning, we have 30 on-campus students and 10 online students signed up for the course.

Tuesday, August 23, 2005

 

AFOSR review in Optimization and Discrete Mathematics

I'm in St. Louis at a review meeting for the Air Force Office of Scientific Research (AFOSR) program in Optimization and Discrete Mathematics. The program may be clicked off this page. In the morning we went to Scott AFB and visited the Air Mobility Command HQ and specifically the Tanker Airlift Command Center. Pretty cool stuff, but if I tell you about it, I have to kill you. For genetic algorithm fans, I ran into Darrell Whitley and David Van Veldhuizen (co-author of one of the most popular books on multiobjective GAs)

Sunday, August 21, 2005

 

Optimatics

IlliGAL Blogging's Paul Winward posted about Optimatics, an Australian firm that uses GAs to optimize water distribution networks. Old time IlliGALers will recall that one of Optimatics principals, Dr. Angus Simpson, was an IlliGAL visitor back in 1993. At the time, Angus studied the use of advanced GAs in water distribution system design, and Optimatics was started shortly after Simpson returned to Adelaide following his Urbana adventure. Angus and I were students together at the University of Michigan, and Ben Wylie advised both of our theses. My own work in water distribution network analysis and design goes back to my four-year stint at Stoner Associates in Carlise, Pennsylvania. At the time Mike Stoner was a lone entrepreneur with a suite of four analysis codes in a new building in the middle of a corn field. Today, Stoner is called Advantica, and the company serves gas and water clients around the globe.

 

Pondering NFL and GAFO

Pensive Pondering is a new blog and the first few posts comment on the De Jong's writing on GAFO (GAs as function optimizers) and Wolpert and Macready's writing on the no free lunch theorem. I've always been a card-carrying member of the field's optimist wing and have almost always found writing about what GAs aren't or can't do (1) trivially true and (2) uninteresting. For me the existence proof of nature's evolved and complex bounty has suggested that what genetic algorithms can do is much more interesting than what they can't. Thus, I believe we should (1) not spend one nanosecond worrying about the NFL limits and (2) not use GAFO and DeJong's thinking as an excuse NOT to explore the limits of how fast and effective GAs can be made. The book The Design of Innovation is my answer to the pessimistic wing of the field and subsequent research continues to tell us how we're just scratching the surface on our understanding of what genetic algorithms can be designed to do.

 

GAs & GP at heart of Humanoid Project

BitKrafted posts about the Chalmers University's Humanoid Project in which layers of genetic algorithms and genetic programming are used as the intelligent core of a full-size humanoid robot. More information is available at the project web page here.

Thursday, August 18, 2005

 

On your mark...get set....

Sodarace is a site where you make stick figures and race them against stick figures that are made by GAs, and other AIs.

Please don't hate me for the hours of your time I've just wasted.

Tuesday, August 16, 2005

 

He's baaaaaack

The Goldberg family returned this weekend from a two-week vacation in Douglas on Michigan's west coast. Douglas is adjacent to Saugatuck, an artsie community with quaint shops, galleries, and access to Lake Michigan down the Kalamazoo River. Between going to the beach, sailing, kayaking, biking, attending lovely summer theatre productions of Urinetown and Honky Tonk Highway at Mason Street Warehouse or traveling to Grand Rapids to go to a Whitecaps minor league ball game, the John Ball Zoo, and the Meijer Gardens, we had a blast, and I think I'm ready to return to the grind of the mind and the slog of the blog. I thank those IlliGAL Bloggers who picked up the slack to bring you the latest in genetic algorithms and evolutionary computation during my little break.

Friday, August 12, 2005

 

AOM 2005

I just came back from the academy of management conference. It's not EC related, but EC and management share some same concepts. For example, modularity plays an important role in both optimization problems in EC and organizational management (for example, teaming). Our current research direction is to apply the concept of little models to the management and organization theory. In the conference, people showed a lot of interests in this concept. I also found several interesting research work there. For example, people have been working on innovation system, managerial cognition, organizational behavior. If you are interested, a full program description is available here. Also, our work entitled "simple models of hierarchical organizations" is available here.

 

GAs for urban water flow

Came across this article about an Australian-based company called Optimatics that uses GAs for water system planning, design, and operations. They suggest that "a cost saving of 20%-30% makes it worthwhile to undertake GA." Apparently, they've done over 50 GA studies in water passage and flow in Australia, North America, and Europe. Each project is highlighted with background info, a problem statement, scope, and results. This can be accessed here. Their recent studies take them to Phoenix, Dayton, and Tulsa and Forsyth County, Georgia.

Wednesday, August 10, 2005

 

Visualization implemented in NSGA2

Recently, I discovered that a new version of NSGA2 from the Kanpur Genetic Algorithms Laboratory KanGAL included a visualization using gnuplot. It is thus possible during the run to track the progress of the algorithm visually. The visualization only works in 2 or 3 dimensions, but it is possible to decide which objectives are shown on the different axes, so even for problems having more than 3 objectives it is possible to choose the objectives that are plotted during the run. I think it is really nice to be able to track the progress of the algorithm during the run, and the way the visualization is implemented using gnuplot is quite good.

Monday, August 08, 2005

 

Deleuze’s intensive thinking and GAs in architecture

I run into another interesting article browsing the net this summer. The Mexican philosopher Manuel DeLanda wrote it. His article explores the connections between different kinds of thinking paradigms (intensive, topological, and population ones, to mention a few) and genetic algorithms when applied to architecture.

 

New Version of Leading Drug Discovery Tool Uses GA

In a press release today, Tripos announced Version 7.1 of its SYBIL Molecular Modeling Tool, which includes a GA, apparently developed in conjunction with researchers in the Chemoinformatics Group of The University of Sheffield. There's a nice presentation on this work here.

Friday, August 05, 2005

 

Frank Gehry, curvilinear design, and genetic algorithms

Browsing the net this summer I run in to an interesting paper by John Middendorf about curvilinear design in architecture. Genetic algorithms play an interesting role in it. Such algorithms help in curvilinear designs such as Frank Gehry’s designs.

Thursday, August 04, 2005

 

GA for diesel engine design

Always on the lookout for GA applications, and came across this somewhat older but interesting article. Peter Senecal, a UW-Madison Ph.D. student, is using GAs to design better diesel engines. The article says this:

"The results to date have been dramatic. Using a silicon graphics supercomputer, Senecal created a diesel engine design that reduces nitric oxide emissions by three-fold and soot emissions by 50 percent over the best available technology. At the same time, the model reduced fuel consumption by 15 percent."

The GA considered fuel injection timing, injection pressure, and amount of exhaust recirculation. Additional information can be found here and here.

This page is powered by Blogger. Isn't yours?